ai-safety-foundation / sparse_autoencoderLinks
Sparse Autoencoder for Mechanistic Interpretability
☆260Updated last year
Alternatives and similar repositories for sparse_autoencoder
Users that are interested in sparse_autoencoder are comparing it to the libraries listed below
Sorting:
- Using sparse coding to find distributed representations used by neural networks.☆264Updated last year
- ☆238Updated 10 months ago
- ☆335Updated last week
- Create feature-centric and prompt-centric visualizations for sparse autoencoders (like those from Anthropic's published research).☆214Updated 8 months ago
- Sparsify transformers with SAEs and transcoders☆609Updated last week
- Mechanistic Interpretability Visualizations using React☆281Updated 8 months ago
- ☆114Updated last month
- ☆126Updated last year
- Delphi was the home of a temple to Phoebus Apollo, which famously had the inscription, 'Know Thyself.' This library lets language models …☆207Updated this week
- ☆184Updated last month
- Training Sparse Autoencoders on Language Models☆925Updated last week
- The nnsight package enables interpreting and manipulating the internals of deep learned models.☆638Updated this week
- ☆163Updated 9 months ago
- ☆513Updated last year
- Steering vectors for transformer language models in Pytorch / Huggingface☆122Updated 6 months ago
- ☆122Updated last year
- Steering Llama 2 with Contrastive Activation Addition☆176Updated last year
- ☆53Updated 9 months ago
- Tools for understanding how transformer predictions are built layer-by-layer☆516Updated 3 weeks ago
- Open source replication of Anthropic's Crosscoders for Model Diffing☆58Updated 10 months ago
- ☆187Updated 9 months ago
- This repository collects all relevant resources about interpretability in LLMs☆369Updated 9 months ago
- Modified to support crosscoder training.☆22Updated last month
- Sparse Autoencoder Training Library☆54Updated 3 months ago
- An Open Source Implementation of Anthropic's Paper: "Towards Monosemanticity: Decomposing Language Models with Dictionary Learning"☆49Updated last year
- ☆81Updated 6 months ago
- For OpenMOSS Mechanistic Interpretability Team's Sparse Autoencoder (SAE) research.☆145Updated last week
- Code and results accompanying the paper "Refusal in Language Models Is Mediated by a Single Direction".☆261Updated 2 months ago
- ☆73Updated 3 months ago
- ☆104Updated 6 months ago